An agent-based opinion dynamics model, incorporating some recent ideas from ANNSIM 2024 and INSNA Sunbelt 2024.
Make sure your Python environment has Mesa, networkx, NumPy, Pandas, and matplotlib installed:
$ pip install mesa networks numpy pandas matplotlib
and then run the single Python file tidal.py, whose command-line usage is
shown below.
tidal.py [-h] [-n NUM_SIMS] [--seed SEED] [--MAX_STEPS MAX_STEPS]
[-N N] [--graph_type {ER,SBM}]
[--graph_params GRAPH_PARAMS [GRAPH_PARAMS ...]]
[--agent_class {Messaging,Community}] [--prop_red PROP_RED]
[--confidence_malleability CONFIDENCE_MALLEABILITY]
[--cap_confidence | --no-cap_confidence]
[--bidirectional_influence | --no-bidirectional_influence]
[--extraversion EXTRAVERSION]
[--animate_only_on_step | --no-animate_only_on_step]
[--plot_mean | --no-plot_mean]
options:
-h, --help show this help message and exit
-n NUM_SIMS, --num_sims NUM_SIMS
Number of sims to run (1 = single, >1 = batch)
--seed SEED Random seed.
--MAX_STEPS MAX_STEPS
Maximum number of steps before simulation terminates.
-N N Number of agents.
--graph_type {ER,SBM}
Random graph-generating algorithm.
--graph_params GRAPH_PARAMS [GRAPH_PARAMS ...]
Random graph-generating algorithm parameters (ER: p.
SBM: n1 n2 p11 p12 p22).
--agent_class {Messaging,Community}
Prefix of agent class name (prepended to 'Citizen').
--prop_red PROP_RED Proportion of agents initially red.
--confidence_malleability CONFIDENCE_MALLEABILITY
To what degree is confidence boosted/devastated by
(dis)agreement.
--cap_confidence, --no-cap_confidence
Impose maximum confidence of 1.0 on each agent?
--bidirectional_influence, --no-bidirectional_influence
Listener also (symmetrically) influences speaker?
--extraversion EXTRAVERSION
Probability of each agent sending message each step.
--animate_only_on_step, --no-animate_only_on_step
Only draw new animation frame on entire step of sim?
--plot_mean, --no-plot_mean
Plot mean and median confidence on histogram?